Since Stolt residual prestack migration is based on the assumption of constant velocity, we will evaluate the accuracy for velocity models having depth dependence.

We create our second synthetic model
using the same reflector geometry as the previous
section and using a realistic depth velocity model
of *v*_{0p}=1700 m/s with a gradient of 0.15 s^{-1}
and *v*_{0s}=300 m/s with a gradient of 0.35 s^{-1},

Figure 5

As before,
we perform the initial shot profile migration with
a positive perturbation
in the *S*-velocity model.
We perform residual migration with the RMS value at approximately 1.5 km,
and and .Residual prestack migration could not recover the image correctly
at all depths.
However, we got a good approximation of it.
Figures 5
and 6 show our results.

Figure 6

9/18/2001